Maize Regional Trial Analysis Models And Methods

Maize is one of the world’s three major crops, China is the second largestcorn-producing country in the world, corn not only occupies a pivotal position in the foodproduction, but also plays an important role in the national economy. in China, corn is mainlyused for feeding, consumption and industrial application, and feeding that reach70percent ofthe total demand occupies the biggest percent. With the rapid development of fuel ethanolindustry, the corn demand for industrial consumption is growing. Therefore, improving maizeproduction, accelerating the development of the corn is particularly important. In order toimprove the production levels of corn, except accelerating the development of new varietiesof corn, and constantly improve the high yield of maize cultivation and managementtechniques, reasonable scientific analysis methods and models of maize regional trials andvariety evaluation are also critical.The lack of systematic comparative study the effects test analysis and the differencesbetween different kinds of methods and models in the crop regional trials particularly in themaize trials affects the reasonable application of the analysis methods in maize regional trial.This limits the effectiveness of the test analysis of the effect on corn breeding and cornproduction. This study based on corn variety of regional test data in recent years, do thesystematic comparative study of the models and methods of the corn test and obtain thefollowing main conclusions:（1）The results of the evaluation and sorting of maize varieties on the analysis ofvariance and the gray relational analysis are obviously different, so it is necessary to choosethe reasonable analytical methods. When analysis a number of ecological regions andanalysis single trait such as yield, we should use the analysis of variance; when analysis asingle ecological region and analysis a number of traits, we should use the gray relationalanalysis.（2）High stability coefficient method and the rank analysis method to evaluate highyield and yield stability of varieties in the crop regional trial, and the high yield results areconsistency, but stability evaluation results have the large difference. Therefore, it isnecessary to choose the reasonable analytical methods when evaluate yield stability ofvarieties.（3）A variety of variance covariance structure of the linear mixed model has different effect on corn test data fitting. Analysis of variance model fit the experimental data poorly,because AMMI and FA (1) structural model relax the constraints of each other differentialcovariance and have less Model parameters, so the two models have the most widelyapplicable. The UN structure model does not apply to the corn regional trial analysis in China,but other structural models apply to corn regional trial analysis in China, and it is worse thanthe AMMI and FA (1) structural model. There is a significantly difference between the varietyeffect estimation and tests through these structures models. Therefore, we should use theAMMI and FA (1) structural model or use the AIC criterion to choose the best model whenuse the linear mixed model to analysis the maize regional trial,.（4）Different stability analysis models have different effects and fitting results in theregional trial. High stability coefficient method and the rank analysis method are twonon-parametric methods, compared to several parametric models, the former difference of thestability is lager than the latter. The AMMI model showed the best effect in the most mazeregion trial, it is the best stability evaluation method. Therefore, In order to accuratelyevaluate the stability of the varieties, in the stability analysis of the corn variety of regionaltrial, we should use the AIC criterion to choose the best model or use the AMMI model toanalysis.（5）Because the interaction effects of variety and environment, it’s not very clear forpeople to know which model and method fit the maize regional trial, especially the effect onhigh stability coefficient method and the rank analysis method, there is not yet a practicalmethod to judge. The linear mixed model analysis can not only estimates the effect ofvarieties and tests stability evaluation, but also choose the best model to analysis according tothe fitting effect of the test data during the models, it’s expected to have a good prospect ofapplication of corn regional trial analysis. But the direct verification of the analysis of theaccuracy of the model needs further simulation studies.